arX

iv:1

206.

1510

v2 [

astr

o-ph

.EP]

4 J

un 2

013

The PTF Orion Project: a Possible Planet Transiting a T-Tauri Star

Julian C. van Eyken,1,17,18 David R. Ciardi,1,18 Kaspar von Braun,1 Stephen R. Kane,1

Peter Plavchan,1 Chad F. Bender,2,3 Timothy M. Brown,4 Justin R. Crepp,5 Benjamin J. Fulton,4

Andrew W. Howard,6 Steve B. Howell,7,18 Suvrath Mahadevan,2,3 Geoffrey W. Marcy,6

Avi Shporer,4 Paula Szkody,8,18 Rachel L. Akeson,1 Charles A. Beichman,1 Andrew F. Boden,9

Dawn M. Gelino,1 D. W. Hoard,10 Solange V. Ramırez,1 Luisa M. Rebull,10 John R. Stauffer,10

Joshua S. Bloom,6 S. Bradley Cenko,6 Mansi M. Kasliwal,5 Shrinivas R. Kulkarni,5

Nicholas M. Law,11 Peter E. Nugent,12 Eran O. Ofek,13,19 Dovi Poznanski,14 Robert M. Quimby,15

Richard Walters,5 Carl J. Grillmair,10 Russ Laher,10 David B. Levitan,16 Branimir Sesar,5

and

Jason A. Surace10

vaneyken@ipac.caltech.edu

– 2 –

ABSTRACT

We report observations of a possible young transiting planet orbiting a previously

known weak-lined T-Tauri star in the 7–10 Myr-old Orion-OB1a/25-Ori region. The

candidate was found as part of the Palomar Transient Factory (PTF) Orion project.

It has a photometric transit period of 0.448413 ± 0.000040 days, and appears in both

2009 and 2010 PTF data. Follow-up low-precision radial velocity observations and

adaptive-optics imaging suggest that the star is not an eclipsing binary, and that it is

unlikely that a background source is blended with the target and mimicking the observed

transit. Radial-velocity observations with the Hobby-Eberly and Keck telescopes yield a

radial velocity that has the same period as the photometric event, but is offset in phase

1NASA Exoplanet Science Institute, California Institute of Technology, 770 South Wilson Avenue, M/S 100-22,

Pasadena, CA, 91125, USA

2Department of Astronomy & Astrophysics, The Pennsylvania State University, University Park, PA 16802, USA

3Center for Exoplanets and Habitable Worlds, The Pennsylvania State University, University Park, PA 16802,

USA

4Las Cumbres Observatory Global Telescope, Goleta, CA 93117, USA

5Cahill Center for Astrophysics, California Institute of Technology, Pasadena, CA, 91125, USA

6Department of Astronomy, University of California, Berkeley, CA 94720-3411, USA

7NASA Ames Research Center, M/S 244-30, Moffett Field, CA 94035, USA

8Department of Astronomy, University of Washington, Box 351580, Seattle, WA 98195, USA

9Caltech Optical Observatories, California Institute of Technology, Pasadena, CA 91125, USA

10Spitzer Science Center, M/S 220-6, California Institute of Technology, Jet Propulsion Laboratory, Pasadena, CA

91125, USA

11Dunlap Institute for Astronomy and Astrophysics, University of Toronto, 50 St. George Street, Toronto M5S

3H4, Ontario, Canada

12Computational Cosmology Center, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA

94720, USA

13Benoziyo Center for Astrophysics, Weizmann Institute of Science, 76100 Rehovot, Israel

14School of Physics and Astronomy, Tel Aviv University, Tel Aviv 69978, Israel

15IPMU, University of Tokyo, Kashiwanoha 5-1-5, Kashiwa-shi, Chiba, Japan

16Division of Physics, Mathematics and Astronomy, California Institute of Technology, Pasadena, CA 91125, USA

17Currently at Department of Physics, UC Santa Barbara, Santa Barbara CA 93106; vaneyken@physics.ucsb.edu

18Visiting Astronomer, Kitt Peak National Observatory, National Optical Astronomy Observatory, which is oper-

ated by the Association of Universities for Research in Astronomy, Inc. (AURA) under cooperative agreement with

the National Science Foundation

19Einstein Fellow

– 3 –

from the transit center by ≈ −0.22 periods. The amplitude (half range) of the radial

velocity variations is 2.4 km s−1 and is comparable with the expected radial velocity

amplitude that stellar spots could induce. The radial velocity curve is likely dominated

by stellar spot modulation and provides an upper limit to the projected companion

mass of Mp sin iorb . 4.8 ± 1.2MJup; when combined with the orbital inclination, iorb,

of the candidate planet from modeling of the transit lightcurve, we find an upper limit

on the mass of the planetary candidate of Mp . 5.5± 1.4MJup. This limit implies that

the planet is orbiting close to, if not inside, its Roche limiting orbital radius, so that it

may be undergoing active mass loss and evaporating.

Subject headings: Open clusters and associations: individual (25 Ori) – planets and

satellites: detection – stars: individual (2MASS J05250755+0134243 / CVSO 30 /

PTFO 8-8695 / PTF1 J052507.55+013424.3) – stars: pre-main sequence

1. Introduction

The Palomar Transient Factory (PTF) Orion project is a study within the broader PTF survey

aimed at searching for photometric variability in the young Orion region, with the primary goal

of finding young extrasolar planets (van Eyken et al. 2011). The project is based on a set of

intensive high-cadence (≈ 70–90 s) observations of a single 7.26 deg2 field centered on the known

25 Ori association, which lies in the Orion OB1a region and has an estimated age of 7–10 Myr

(Briceno et al. 2005, 2007).

Typical young circumstellar disk lifetimes are on the order of 5–10 Myr (Hillenbrand 2008), and

it is during this time that the bulk of the formation and migration of planets is expected to occur.

The youngest exoplanets have been found via direct detection (e.g., LkCa 15, Kraus & Ireland

2012; the free-floating planetary-mass object in the ρ Oph cloud, Marsh et al. 2010); however,

little is known empirically about exoplanets during the first few millions of years of their lives,

and the goal is to fill the observational gap and begin to provide constraints on theories of planet

formation and evolution (see, e.g., Armitage 2009). The stars in the 25 Ori/OB1a region should

be at or just beyond the point of disk dissipation, consistent with the high ratio of weak-lined to

classical T-Tauri stars found there (Briceno et al. 2007). We can therefore look for planet transits

at the time when they may first become observable without their signatures being swamped by

the extreme variability characteristic of the younger classical T-Tauri stars. In so doing, we can

begin to investigate the frequency of planets at these ages; the timescales for their evolution; the

timescales of their migration with respect to the star’s evolution; and, through measurements of

the transit depths, probe empirically their mean densities and the extent of their atmospheres,

which are expected to be inflated at these early stages (Fortney & Nettelmann 2010, and references

therein).

Several surveys have been undertaken to search for close-in young exoplanets, though many

– 4 –

are radial-velocity searches (e.g., Esposito et al. 2006; Paulson & Yelda 2006; Setiawan et al. 2007,

2008; Huerta et al. 2008; Crockett et al. 2011; Nguyen et al. 2012). A transit search (e.g., Aigrain et al.

2007; Miller et al. 2008; Neuhauser et al. 2011) has the advantage of being able to search many more

stars simultaneously, down to fainter magnitudes and with a faster cadence. Combining transit pho-

tometry with spectroscopic information can provide a wealth of information that is unavailable with

non-transiting planets, from masses and radii to constraints on atmospheric composition.

An outline of the PTF Orion project is given by van Eyken et al. (2011), along with some of

the first results concerning binary stars and T-Tauri stars. The broader PTF survey is described in

detail by Law et al. (2009) and Rau et al. (2009), with a summary of the first year’s performance

by Law et al. (2010). Here we report a young planet candidate found orbiting a known M3 pre-

main-sequence (PMS) weak-lined T-Tauri star (TTS) within the PTF Orion field. Our observations

are outlined in Section 2; Section 3 discusses the photometric and spectroscopic results and some

implications. The main conclusions are summarized in section 4.

2. Observations

2.1. PTF Photometry

PTF Orion data were obtained using the Palomar 48′′ Samuel Oschin telescope during the

majority of the clear nights between 2009 December 1 and 2010 January 15, whenever the field was

above an airmass of 2.0. All observations were in the R band, and of the 40 nights dedicated, 14

yielded usable data, the remainder being lost primarily to poor weather. Light curves were obtained

for ∼ 110, 000 sources within the field, the top ∼ 500 most variable of which were inspected visually.

The observations and the PTF Orion differential photometry pipeline and data reduction details

are described fully in van Eyken et al. (2011). Specifics of the more general PTF data reduction

pipeline are described by R. Laher et al. (2012, in preparation) and Ofek et al. (2012). Among

the inspected sources, one, PTFO 8-8695 (2MASS J05250755+0134243), showed periodic shallow

transit-like events with a shape and depth suggestive of a planetary companion, superposed on

a larger-scale quasi-periodic variable light curve. The source has previously been identified as a

weak-line T-Tauri star (WTTS) associated with the Orion OB1a region by Briceno et al. (CVSO

30, 2005), and Hernandez et al. (2007). A summary of the main previously determined properties

of the primary star is given in Table 1. In Figure 1 we show a color-magnitude diagram indicating

PTFO 8-8695 in relation to the other T-Tauri stars in the OB1a region discovered by Briceno et al.

(2005). Although classified as a WTTS, it lies at the younger end of the distribution. This is

consistent with its its relatively strong Hα emission, which in fact places it on the borderline with

classical T-Tauri stars (CTTS) according to the classification scheme of Briceno et al. (2005).

The same field was again observed in the same way over seven clear nights between 2010

December 8–17. The transit events were again evident with the same period and depth. Owing to

improvements in the PTF image processing software, better weather, and mitigation of the CCD

– 5 –

Table 1. PTFO 8-8695 Stellar Properties

Property Value

Alternative designations CVSO 30

2MASS J05250755+0134243

PTF1 J052507.55+013424.3

α (J2000) 05.h25.m07.s55

δ (J2000) +01.◦34.′24.3.′′

V 16.26 maga

2MASS J 12.232 ± 0.028 magb

2MASS H 11.559 ± 0.026 magb

2MASS KS 11.357 ± 0.021 magb

Median R 15.19 magc

R range 0.17 mag (min to max)c

Hα equiv. width -11.40Aa

LiI equiv. width 0.40 Aa

Sp. Type M3 (PMS weak-lined T-Tauri)a

Teff 3470 Ka

AV 0.12 maga

Luminosity 0.25L⊙a

Radius 1.39R⊙a,d

Mass (Baraffe/Siess) 0.44M⊙/0.34M⊙a,e

Age (Baraffe/Siess) 2.63 Myr/2.68 Myra,e

Distance ∼ 330 pc (mean dist. to OB1a/25 Ori assoc.)a,f

aBriceno et al. (2005).

bSkrutskie et al. (2006).

cFrom PTF Orion data (this paper).

dcf. smaller value implied by PTF Orion transit measurements – see Table 3.

eReported by Briceno et al. (2005) based on comparison with Baraffe et al. (1998) and Siess et al.

(2000) stellar models.

fBriceno et al. (2007)

– 6 –

Fig. 1.— Color-magnitude diagram following Briceno et al. (Figure 7, 2005), highlighting PTFO

8-8695 (CVSO 30) – the star symbol – in relation to the other T-Tauri stars discovered therein (see

section 2.1). PTFO 8-8695 lies at the younger end of the distribution. Open circles indicate WTTS;

filled circles indicate CTTS. Solid lines indicate, from the top, 1, 3, 10, and 30 Myr isochrones, and

the zero-age main sequence, according to the models of Siess et al. (2000) at a distance of 330 pc.

Photometric measurements are reproduced from Briceno et al. (2005). Reddening and extinction

are neglected.

– 7 –

fogging effect seen in the previous year’s data (see van Eyken et al. 2011; Ofek et al. 2012), the RMS

noise floor for the brightest stars in the 2010 December data set improved from ≈ 4 to ≈ 3 mmag,

with less evidence of systematic effects.1 Figures 2 and 3 show the light curves obtained in the first

and second years of observations with PTF. The quasi-periodic stellar variability is evident, along

with sporadic low-level flaring, consistent with the star’s young age and late spectral type. Visual

inspection of the curves revealed the additional regular periodic transit-like signature, with a depth

of ≈ 3–4%, superposed on the stellar variability. We initially determined an approximate period

(≈ 0.45 days) and transit duration (≈ 2 hours) by hand, in order to locate all the transit windows. A

detailed discussion of the light curves and derived properties is given in Section 3.1. The regularity

and planet-like characteristics of the transit events led us to obtain follow-up observations.

2.2. Keck Adaptive Optic Imaging

The PTF imaging instrument is seeing limited and has a point-spread function (PSF) with

typical full-width-half-maximum (FWHM) of 2.0′′, or 660 AU at the estimated distance of the Orion

OB1a and 25 Ori associations (330 pc, Briceno et al. 2005, 2007). We obtained adaptive optics (AO)

images using the NIRC2 camera (PI K. Matthews) on the 10 m Keck II telescope in order to probe

regions in the immediate vicinity of the star and to search for any sources (false positives) that

could mimic the signal of a transiting planet, such as a nearby eclipsing binary. At R = 15.2 mag,

PTFO 8-8695 is sufficiently bright for natural guide star observations and does not require the

use of a laser for atmospheric compensation. We locked the AO system control loops onto the

target using a frame rate of 41 Hz. The airmass was 1.29. We used the NIRC2 narrow camera

setting to provide fine spatial sampling (10 mas pix−1). Our observations consisted of 12 dithered

H -band images (3 coadds per frame, 10 s per coadd), totaling 6 minutes of on-source integration

time. Raw frames were processed by cleaning hot pixels, subtracting background noise from the

sky and detector, and aligning and coadding the results.

Figure 4 shows the final reduced image using an effective field-of-view of 4.9′′ × 4.9′′, which

corresponds to ≈ 4.9 PTF pixels on a side. The image shows no contaminants, except for one faint

source to the south east, at a separation of 1.8′′ (590 AU in the plane of the sky at the distance

of Orion OB1a/25 Ori). Our diffraction-limited images (FWHM ≈ 80 mas) rule out additional

off-axis sources down to a level of ∆H = 4.3, 6.4, 8.9, and 9.1 mag (3σ) at angular separations of

0.25′′, 0.5′′, 1.0′′, and 2.0′′ (83, 165, 330, and 660 pc) respectively. Assuming a color difference of

0 mag, a faint, blended, 100% eclipsing binary would have to be within ∆H ≈ 3.5–3.8 mag of the

primary star, or brighter, to mimic the observed transit depth at R of ≈ 3–4%. At 6.96 mag (608

times) fainter than our target, the one imaged contaminant is too faint to be a blended eclipsing

binary capable of mimicking the observed transits unless it is extremely blue (R −H . −3.2); in

1Independent differential photometric analyses of other PTF data, and respective precision levels, are discussed

in Agueros et al. (2011) and Levitan et al. (2011).

– 8 –

Fig. 2.— Photometric light curve for 2009 December 1–2010 January 15 (Sections 2.1 and 3.1.1).

Gray regions indicate the transit windows, fixed at the measured transit period, width, and epoch

of center-transit (T0). Red points indicate data automatically flagged by the data reduction soft-

ware as potentially non-optimal for various possible reasons (e.g., imperfect weather, evidence of

contamination within the photometric aperture, etc.)

– 9 –

Fig. 3.— As for Figure 2, for 2010 December 8–17.

– 10 –

Fig. 4.— Adaptive optics imaging of PTFO 8-8695, showing the full image field (left), and detail

around the central star (right), at two different logarithmic brightness stretches. A very faint source

is detected to the south east of the central star at a separation of 1.8′′, 6.96 mag fainter than the

central source, and unlikely to be capable of mimicking the observed transits (indicated by the

arrow, left panel); otherwise no contaminants are detected. (Section 2.2.)

that event, however, it would be unlikely to be stellar in origin in any case. Any other such binary

would have to lie within 0.25′′ of our target in order not to have been detected.

2.3. LCOGT, KPNO 4m, and Palomar 200′′ Vetting

Photometric observations were obtained in 2011 February with the Las Cumbres Observatory

Global Telescope Network (LCOGT), using the 2 m Faulkes North and South telescopes, and the

0.8 m Byrne Observatory telescope at Sedgwick Reserve (BOS). We were able to confirm complete

transit detections on two separate nights with the BOS telescope with a clear filter (see Figure 5).

Observations in SDSS g′ and i′ with the three telescopes were inconclusive, and largely hampered

by poor observing conditions and sparse phase coverage. The LCOGT observations enabled us to

confirm the transit event and its period independently of the PTF data, and, in combination with

the PTF data, we were able to establish a more accurate and up-to-date transit ephemeris.

In addition we obtained low-precision followup radial velocity (RV) observations with the

KPNO 4 m Mayall telescope and the Palomar 200′′ Hale telescope over 3 and 2 consecutive nights,

respectively, to allow us to rule out a stellar-mass companion as the cause of the transits. The

separate confirmation of the transit event and the lack of RV signature at the level of ∼ 10 km s−1

suggested that we had indeed detected a sub-stellar object, and we pursued additional observations.

– 11 –

Fig. 5.— Transit observations obtained with the LCOGT BOS 0.8 m telescope (clear filter), on

the nights of 2011 February 8 and 9 (Section 2.3). Gray regions again indicate the predicted

transit windows using the period and T0 obtained with the PTF data. The second night shows a

possible small flare at the end of the transit. Note that in addition to stellar variability, some of

the systematic trends may also be air-mass related.

2.4. HET and Keck Spectroscopy

Following the AO vetting and low-precision RV followup, we obtained Doppler radial ve-

locity spectroscopy with both the High Resolution Spectrograph (HRS; Tull 1998) on the 9.2 m

Hobby Eberly Telescope (HET; Ramsey et al. 1998), and the High Resolution Echelle Spectrom-

eter (HIRES; Vogt et al. 1994) on the 10 m Keck I telescope. The goal was to detect or place an

upper limit on any signal due to reflex stellar motion caused by the orbiting companion.

The queue-schedule operation mode of the HET (Shetrone et al. 2007) allowed us to obtain

four spectroscopic radial velocity observations very quickly after analysis of the low-precision RV

and LCOGT photometric vetting data. The observations were timed on the basis of prior knowledge

of the transit ephemeris to best constrain any orbital signature.

Observations from the fiber-fed HRS spectrograph and calibrated using ThAr exposures pro-

vide radial velocity precision of better than 50 m s−1 over timescales of several weeks (Bender et al.

2012). Seeing-induced PSF changes are minimal due to the image scrambling properties of optical

fibers (Heacox 1987) and the HRS temperature is kept stable to ∼ 0.01 C. We used the 15k reso-

lution mode of HRS, with the red cross-disperser setting (316g7940), giving a wavelength coverage

of 6114–9861 A. The use of this lower resolution setting enhances the efficiency of HRS without

significantly degrading the information content in the stellar spectrum (since the target absorption

lines were already known to be rotationally broadened). Each observation was 1200 s in duration,

except on UT 2011-02-21 which was 1600 s, and ThAr calibration frames were taken immediately

after each to track instrument drift. Data for the four nights were reduced in IDL with a custom op-

timal extraction pipeline that performs bias subtraction, flat fielding, order tracing and extraction,

and wavelength calibration of the extracted one-dimensional spectra. A simultaneous sky-fiber was

– 12 –

available with this spectrograph setting, but we chose to avoid the additional complexity inherent

in performing sky subtraction with a fiber-fed spectrograph, and instead masked areas affected by

sky lines so that they were not used in calculating the RVs. The achieved signal to noise in each

spectrum was approximately 40–50 per resolution element, peaking in the I -band. We identified

the strongest absorption lines in an M2-type high resolution Kurucz stellar model2 (Kurucz 2005),

and used these to select lines for fitting in the measured spectra.

For the Keck data, we observed PTFO 8-8695 using the standard procedures (e.g., Howard et al.

2009) of the California Planet Search (CPS) for HIRES. The five observations over 10 days were

each 500–600 s in duration and achieved a signal to noise ratio of 10–20 per pixel (S/N = 20–40

per resolution element) in the V -band. We used the “C2” decker (0.86′′ wide slit) for a spectral

resolution of ∼60,000. The total wavelength coverage was ≈ 3650–7970 A.

Owing to the faintness of the target and the relatively relaxed precision requirements, we

employed in both cases thorium-argon (ThAr) emission lamps as the fiducial reference. This avoids

the light-throughput penalty incurred by the higher precision common-path iodine gas absorption

cell, which is better suited to brighter targets.

Data reduction and analysis were performed independently at separate institutions for the two

data sets, using independently written software, but following the same general principles. Visual

inspection of the spectra readily revealed strong rotational line broadening (vr sin i∗ = 80.6 ±

8.1 km s−1 – see section 3.2.2), to the extent that only the very strongest of the photospheric lines

were evident above the noise level. The low signal-to-noise ration (S/N) and high stellar rotational

velocity resulted in a very small number of lines from which to measure the radial velocity. Using

a cross-correlation analysis on the HRS spectra to measure these broad, low S/N lines yielded a

velocity uncertainty of ∼ 10 – 14 km s−1. Such an analysis is very sensitive to bad pixels or faint sky

lines, which were difficult to remove in the low S/N regime. In addition, a cross correlation strongly

benefits by combining the information content of multiple lines with fixed relative spacings. In our

data, viable lines tended to be spaced far apart: typically there were only one or two lines per

spectral order, which negated much of the advantage of using a cross-correlation. Consequently, we

adopted an alternative approach of fitting individual stellar lines with Gaussian functions, solving

each line independently for its center, width, and depth. This fitting procedure was much less

sensitive than cross-correlation to bad pixels or to the precise spectral window being analyzed. It

also allowed us to mask out telluric absorption and strong, narrow night sky emission lines.

The faintness of the target and the small semi-major axis of the putative companion’s orbit

meant that very high RV precision was neither anticipated nor required. Simple Gaussian+linear-

term profiles provided an adequate fit (i.e., reduced χ2 ∼ 1) owing to the low S/N and heavy

broadening (and probable blending) of the lines, where more sophisticated models would have

added unnecessary and unconstrained extra parameters. A weighted average of the shifts in the

2http://kurucz.harvard.edu

– 13 –

fit centroids with respect to the rest wavelengths of the stellar lines provided the required Doppler

shifts. Since substantially differing S/N in the various lines precluded a simple standard-deviation–

based estimate of the measurement errors, errors were instead estimated by propagating the formal

errors from the Gaussian fits to the lines. In the case of the Keck data, corrections were made for

measured variation in the telluric lines to account for residual uncalibrated instrument drift, but

these corrections were found to be negligible. A total of 14 absorption features spanning ∼ 5300–

7700 A were fit in the Keck data, and 14 spanning ∼ 6100–8700 A in the HET data. The measured

RVs are listed in table 2.

3. Discussion

3.1. Photometry

3.1.1. Light Curves and Periodicities

In order to model the transit events for a derivation of the transit and planetary candidate

properties, the effects of the stellar variability in the light curves need to minimized – i.e., the light

curves outside of transit needed to be whitened. After removing data points which are flagged or

have large measurement errors, we fit a smooth cubic spline to all the PTF data which fall outside

the transit windows (using the IDL imsl cssmooth function3), interpolating across the windows

themselves. The fit is then subtracted (in magnitude space) from the entire light curve. Since

most of the stellar variability occurs on longer timescales than the transits (with the exception

of the occasional flares), this process yields the “whitened” light curve that has the majority of

the stellar variability removed, leaving only the transits. Using a standalone version of the NASA

Exoplanet Science Institute (NExScI) periodogram tool,4 we calculate the Plavchan periodogram

(Plavchan et al. 2008) of the combined whitened data sets, to provide a more accurate formal

transit period measurement. We find a clear peak at 0.448413 ± 0.000040 d.5

Figure 6 shows the whitened data for the nights where any in-transit data were obtained, folded

on the measured period. Short-term stellar variability is not corrected, and is probably the most

likely explanation for those transits (particularly those from JDs 2455175, 2455192, and 2455545)

which deviate significantly in shape from the general form of the others. This may be caused by

a combination of low-level flaring, residual disk occultation, and the companion transiting small-

scale stellar surface features. JD 2455192 appears to show a flare immediately after the transit,

3From the IDL Advanced Math & Stats module; see http://www.ittvis.com/idl

4See http://exoplanetarchive.ipac.caltech.edu

5We make the assumption that there is no phase shift in the transit timing between the first and second years,

such that the datasets can be combined to give a year-long time baseline, and thus a very precise period estimate. If

we disregard this assumption, the first year’s data alone gives the best period estimate, P = 0.4486 ± 0.0010 d.

– 14 –

and also likely the tail of a second flare mid-transit, though the gaps in the data prevent a clear

interpretation. JD 2455545 appears to show an early transit egress; however, the mid-transit

variation and the disparity in comparison to other transits in the same year are suggestive of the

above mentioned variability effects, which may confuse and mask the true egress time.

The original “un-whitened” light curve is dominated by stellar variability (see Figures 2 and 3).

If we assume that the large-scale variability is caused primarily by spot modulation, and that the

relative effect of the transit events is negligible, we can use the un-whitened data to investigate the

stellar rotation. Lomb-Scargle periodograms (Lomb 1976; Scargle 1982) of the two years’ data are

shown in Figure 7. Where the Plavchan periodogram is designed for finding regular periodic features

of arbitrary but constant shape (such as a transit), a Lomb-Scargle periodogram is better suited

for finding periodic modes in a more complex signal such as the quasi-periodic stellar variation in

our data: here, the signal appears not to repeat in an exact fashion and therefore does not fold

well at any period (likely because of phase shifts due to changing spot features).

A strong peak is found at 0.4481 ± 0.0022 d, in fact matching well with the transit period.

Another peak is seen at 0.9985 ± 0.0061 d, corresponding closely to the sidereal or solar Earth day,

and therefore most likely an artifact resulting from the observing cadence. The other peaks all

appear to be aliases of these two periods. To confirm this, we modeled the data by creating an

artificial light curve from two summed sinusoids with the same two periods, providing power at the

frequencies of these two peaks. We superimposed artificial transits modeled as a simple inverted

top-hat function, with the same depth, width, and ephemeris as the real transits. Allowing the

phases and relative amplitude of the 0.4481 d and 0.9985 d signals to vary as free parameters, and

requiring that the total amplitude of the light curve remain approximately the same as that of

the actual data, we performed a least-squares fit to the real periodogram to assess how well its

structure could be reproduced. Figure 7 shows the results, with a good match between model and

data. Repeating the test with the 0.4481 d sinusoid omitted (i.e., summing only the one-day signal

and the transits) gave a poor fit, implying that the form of the periodogram cannot be explained

by the transits alone. Since there is clearly strong correlated out-of-transit variability associated

with the star, and there are no other fundamental periods evident in the periodogram, the 0.4481 d

signal appears to be the only likely period for the star.

There are three other notable aliases of the presumed stellar rotation period, P∗, at 0.3092 ±

0.0010 d, 0.8126 ± 0.0070 d, and 4.43 ± 0.31 d (identified in the figure). The larger of these is

substantially above the upper period limit implied by the measured vr sin i∗ of the star (Section

3.2.2), P∗ < 2πR∗/(vr sin i∗) = 0.671 ± 0.092 d. Though it is unlikely that the star would be

coincidentally rotating at an alias of the transit period with the Earth’s rotation, we repeated the

modeling experiment with the model stellar rotation modified to match the remaining two aliases

to ensure that none could in fact be the true fundamental stellar period (see Dawson & Fabrycky

2010). Similar periodograms were obtained, with peaks at similar locations but with differing

strength ratios; none were as good a fit to the periodograms based on the real data. We therefore

conclude that the star is co-rotating or near co-rotating with the companion orbit.

– 15 –

Fig. 6.— Whitened light curves for nights where in-transit data were obtained, folded on the transit

period (Section 3.1.1; left and right panels show first and second year’s data respectively). Flux

values are normalized to unity outside of eclipse, and nights are offset vertically in increments of 0.1

for clarity. Transits are further distinguished in alternating black and gray, and crosses (red in the

online color version) indicate data flagged as potentially compromised. The Julian day on which

each transit occurred is indicated for comparison with Figures 2 and 3. The light gray regions

indicate the transit windows. Note that partially covered transits with little or no data on one

or other side of the transit window are likely to suffer from poor stellar-variability correction and

show significant systematic error.

– 16 –

Fig. 7.— Lomb-Scargle periodogram of the non-whitened light curves from the two years of PTF

Orion observations, compared to simple synthetic models (Section 3.1.1). The presumed stellar

rotation period is marked with a thick arrow (red in the online color version); a second distinct

one-day period, which we attribute to the observing cadence, is marked with a thin (blue) arrow.

The more prominent aliases of these periods are indicated with vertical lines: thick (red) marks

indicate peaks which are predominantly produced by aliases of the presumed stellar rotation period;

thin (blue) marks indicate peaks predominantly produced by aliases of the one-day period. Dotted

(magenta) lines indicate peaks produced by a combination of the two.

– 17 –

We note that it is difficult to model the observed transit events with star spots alone, partic-

ularly in the 2009 December/2010 January data: the short transit duration relative to the period,

the flat bottom of the transits, and the sharp ingress and egress, are all more characteristic of a

transiting object, whereas spots tend to cause smoother, more sinusoidal features as their projected

area changes as they rotate across the stellar disk. It is more likely that we are seeing in the total

lightcurve the effects of both star spots and a transiting object.

3.1.2. Transit Fitting

We fit transit models to the whitened light curves using the IDL-based Transit Analysis

Package (TAP; Gazak et al. 2012). TAP employs Bayesian Markov Chain Monte Carlo (MCMC)

techniques to explore the fitting parameter space based on the analytic transit light curve mod-

els of Mandel & Agol (2002). The package incorporates white- and red-noise parameterization

(Carter & Winn 2009), allowing for robust estimates of parameter uncertainty distributions. We

fit only the transits from nights where a complete transit was observed with adequate coverage

both before and after the transit window, since on these nights the spline fit to remove the stellar

variability is reasonably constrained on both sides of the transit; on nights where there is partial

transit coverage, the spline fit tends to diverge during transit time. We also reject the transits men-

tioned in section 3.1 where stellar variability appears to strongly affect the shape of the transit, and

in addition, the transit from JD 2455544, where the spline subtraction was particularly uncertain

and rather sensitive to the tightness of the spline fit. Thus we fit four of the transits from the first

year’s data (JDs 2455201, 2455202, 2455205, and 2455211), and three from the second year (JDs

2455539, 2455540, and 2455543). The folded data are shown in Figure 8.

For the transit fitting, we hold the orbital period, P , fixed to the transit period determined

above. The eccentricity, e, is fixed at zero, since it is difficult to constrain in the absence of a

secondary eclipse. White and red noise levels are allowed to float separately for each individual

transit, as are linear airmass trends. The remaining parameters are allowed to float, but are tied

across all transits. They include: the epoch of transit center, T0 (we assume there are no transit

timing drifts between the two years); the orbital inclination, iorb; a/R∗, where a is the orbital semi-

major axis, and R∗ is the radius of the primary; Rp/R∗, where Rp is the companion radius; and the

linear and quadratic limb-darkening coefficients for the primary star. In addition, we constrain the

fitting such that Rp/R∗ < 1, and white and red noise components are less than 4% (the depth of the

transit). Parameters are calculated by fitting Gaussians to the dominant peak in the probability

density distributions for each parameter resulting from the MCMC analysis. The center location

of the Gaussian is taken as the parameter value, and measurement errors are estimated as the

dispersion of the fit. We find that the limb-darkening coefficients are largely unconstrained by the

data. The fit is shown overlaid on the folded data in Figure 8; the corresponding parameters are

listed in Table 3.

There is significant variation in the light curve from transit to transit, as can be seen in Figure

– 18 –

Table 2. Differential Radial Velocity Measurements

HJD RV Telescope

( km s−1)

2455613.668022 1.81±0.64 HET

2455615.649694 2.41±0.64 HET

2455616.640274 -2.36±0.64 HET

2455623.622875 0.55±0.64 HET

2455663.744361 -1.53±0.91 Keck

2455670.747785 -0.01±0.96 Keck

2455671.755651 -1.38±1.03 Keck

2455672.757479 0.10±1.02 Keck

2455673.770857 2.83±1.33 Keck

Note. — RV uncertainties listed are formal uncertainties, and do not account for systematic

effects. The offset between the two data sets is arbitrary, and set by shifting each set so that its

mean is zero, excluding one outlier data point – see Sections 2.4 and 3.2.1.

Fig. 8.— Folded light curve for the two years of observations combined, after removing stellar

variability (Section 3.1.1). The first year’s data (2009 December – 2010 January) are plotted

in black; the second year’s (2010 December) are plotted in orange. A change in transit shape

between the two years is evident. The best transit fit to both years combined is over-plotted on the

assumption that Rp < R∗ (Section 3.1.2). Limb darkening is neglected, and error bars are omitted

for clarity; the median photometric error is 0.0046, with an out-of-eclipse standard deviation of

0.0056.

– 19 –

Table 3. System Parameters

Parameter Value

Measured

P 0.448413 ± 0.000040 d

iorb 61.8 ± 3.7◦

a/R∗ 1.685 ± 0.064

Rp/R∗ 0.1838 ± 0.0097

T0 (HJD) 2455543.9402 ± 0.0008

vr sin i∗ 80.6 ± 8.1 km s−1

Derived

a 0.00838 ± 0.00072 AU

= 1.80 ± 0.15R⊙a

R∗ 1.07 ± 0.10R⊙

Rp 1.91 ± 0.21RJup

Mp sin iorb ≤ 4.8 ± 1.2MJupb

Mp ≤ 5.5 ± 1.4MJup

Note. — Summary of parameters determined in this paper for the PTFO 8-8695 system. Quan-

tities are P – orbital period; iorb – orbital inclination; i∗ – inclination of stellar rotation axis; a –

orbital semi-major axis; R∗ – stellar radius; Rp – planet radius; T0 – epoch of transit center; vr –

stellar equatorial rotational velocity; Mp – planet mass.

aFrom Kepler’s third law, assuming stellar mass M∗ = 0.39± 0.10M⊙ (Briceno et al. 2005), and

Mp ≪ M∗.

bUpper limit derived from measured RV semi-amplitude.

– 20 –

6. This can reasonably be attributed to the companion passing across varying surface features –

cold or hot spots, or perhaps flares – on the stellar photosphere. Such variations are a result of the

planet tracing the stellar surface brightness profile as it traverses the stellar disk, and cannot be

removed by the whitening process, which is only sensitive to the integrated brightness of the disk.

Since PTFO 8-8695 is expected to be active and spotted, such variation in the transits is further

suggestive of a genuine transit, rather than a background blend.

It can also clearly be seen from Figures 6 and 8, however, that there is an overall change in

the transit shape between the two years’ data sets. Explanations for this remain speculative. Re-

running the fitting process and allowing a change in both R∗ and Rp between the two years yields a

decrease in stellar radius of ≈ 10% (2.1σ) from one year to the next (with no measurable change in

companion radius). Such a large change in the stellar radius in such a short time period is unlikely,

and would likely manifest itself in observable rotation rate changes not seen in the data. A change in

transit shape could also arise in principle from a change in orbital geometry. For example, the host

star is rotating quickly enough to exhibit significant oblateness, which may induce precession of the

orbital plane and, therefore, changes in iorb if the orbital and stellar rotational axes are misaligned.6

An additional planet in a different orbit could also produce a similar effect (Miralda-Escude 2002).

It is difficult, however, to explain how an inclination change can yield a transit that becomes

both more grazing (longer ingress/egress) and deeper at the same time. Another explanation may

be variation in star spot coverage (or limb-darkening, though with heavy spot coverage, the two

effects become somewhat confused). In addition to short-term spot variations, there may be surface

features which survive for much longer periods (see, e.g., Mahmud et al. 2011). Given the activity

of the star and the extreme proximity of the companion, such features may be compounded by

magnetic or tidal star/planet interactions that could give rise to varying hot or cold spots near the

sub-planetary point on the stellar photosphere. This could affect the apparent shape of the transit

in a systematic way. The planet’s apparent proximity to the tidal disruption limit (see section 3.3)

may also be a factor: a tidally distorted shape, or transient rings or a tidal tail of evaporating

material could all yield unexpected and possibly varying transit shape (and also cause the slight

transit asymmetry seen in the second year’s data). For the sake of simplicity, and given the lack of

data to disentangle all of these possibilities, we here adopt the single combined fit to both years’

data sets, although with the caveat that the variability may cause some systematic error in the

measurements.

We note that our fit yields a somewhat smaller stellar radius, R∗ ≈ 1.07R⊙, than that pre-

6Given the small orbital period, this may occur on short timescales. Following Miralda-Escude (2002), we estimate

the order of magnitude of the oblateness-induced gravitational quadrupole moment of the host star, J2, by scaling

to the pre-main–sequence from a solar value of ∼ 10−7–10−6 according to J2 ∝ R3∗/(M∗P

2∗). Assuming that the

stellar obliquity is small, and noting that the orbital angular momentum and stellar rotational angular momentum

are similar for Mp ∼ 5MJup (so that the inclinations of the stellar rotation and orbital planes with respect to the

mean plane are similar), this leads to a precession period of the orbital nodes on the order of tens to hundreds of

days.

– 21 –

viously reported by Briceno et al. (2005) (1.39R⊙). Assuming their estimate of Teff = 3470 K

is correct, interpolating Siess models (Siess et al. 2000) with this updated radius gives a slightly

older age estimate for the star, ≈ 3.7 Myr vs. 2.7 Myr. Given the distance uncertainties in the

Briceno et al. (2005) results, however, and the possibility of uncertainty in Teff arising from heavy

spotting in the stellar photosphere, the two radius estimates are probably not inconsistent.7

3.2. Spectroscopy

3.2.1. Radial Velocities

Since the RV analysis is differential in nature, the RV offset between the HET and Keck

data sets is arbitrary. We place them on the same approximate scale by shifting the RV zero

points to the mean of the data for each data set. There are too few data points to create a

periodogram to measure any periodicities in the data, but we can look for consistency with the

previously determined transit period by folding the data on that period and looking for a coherent

alignment of the data points. The result is shown in Figure 9: indeed, the data appear to phase

well, showing a smooth and apparently sinusoidal variation, with the exception of one outlying

data point at orbital phase φ = −0.082 (Heliocentric Julian date (HJD) 2455615.64969), which is

discussed further below. This outlier is excluded in calculating the offset between the data sets.

To constrain the mass of the companion, we fit a Keplerian orbit model to the data using the

RVLIN package by Wright & Howard (2009). The model includes six parameters: the period, P ;

the RV semi-amplitude, K; the eccentricity, e; the argument of periastron, ω; the time of periastron

passage, Tp; and the systemic velocity, γ. We fixed P and Tp to the values measured from the transit

photometry. Since the χ2 surface has multiple minima, we explore the fitting parameter space by

running 10000 trial Keplerian fits (neglecting the apparent outlier point), with initial parameter

estimates drawn at random from reasonable starting distributions, and selecting the fit with the

lowest reduced χ2 (χ2red).

We find that the constraints from the transit ephemeris make it difficult to obtain a reasonable

Keplerian fit. A good circular-orbit fit, with the eccentricity, e, fixed to zero, is prohibited by the

constraint on T0 provided by the photometry. Such a fit must cross its central velocity at phases 0

and 0.5, with minima and maxima at 0.25 and 0.75 (the quadrature points), and as a result appears

significantly out of phase with the data (χ2red = 4.0; dashed line, Figure 9).

An eccentric fit goes some way to resolving the mismatch (χ2red = 1.1), but yields an eccentricity

of e ≈ 0.492 (dotted line, Figure 9). This places the fractional periastron distance at rperi/a ≡

7Our radius estimate also depends on our estimate of a, which depends in turn on the assumption that the mass

estimate of Briceno et al. (2005) is correct, so the argument is somewhat circular. a, however, is relatively insensitive

to stellar mass (a ∝ M1/3), and the Siess models predict a negligible difference in mass at our older age. (In fact, at

the estimated Teff , they predict a mass range smaller than the errors over an age range of 1–10Myr.)

– 22 –

Fig. 9.— Differential radial velocity measurements obtained with Keck/HIRES and HET/HRS

(Section 3.2.1). Zero phase is chosen to correspond to the photometric center-of-transit time. The

offset between the two datasets is chosen to match the mean RV of each. The lines indicate best

Keplerian fits (excluding the outlier): dashed line – circular orbit, transit-center time, T0, fixed

to photometry (χ2red = 4.0); dotted line – eccentric orbit, transit-center time fixed to photometry

(χ2red = 1.1); solid line – a sinusoidal fit (equivalent to a circular orbit) at the same period, with

phase free to float (χ2red = 0.42). The eccentric fit is better than the fixed-T0 fit, but brings the

companion to the surface of the star at periastron. The floating-phase sinusoidal fit gives the best

result, suggesting that star spots either modify or dominate the Doppler RV signal. The outlier

point may represent Rossiter McLaughlin effect due to the transiting companion.

– 23 –

1 − e ≈ 0.51, bringing the companion to the point of contact with the surface of the star as

determined from the photometric transit fits (R∗/a ≈ 0.51). Such a solution may be improbable,

given that in section 3.3 we note that the orbital semi-major axis appears to be close to the Roche

limit.

A better fit is in fact obtained by fitting a circular orbit model and removing the constraint

on T0, allowing the phase to float. This fit is over-plotted in Figure 9, showing the fit is good

(χ2red = 0.42), but offset in phase from the transit ephemeris by −0.22 ± 0.04 periods.

We favor this better fit to the RV signal, which most likely arises because of spot effects

modulated by the stellar rotation, where the amplitude of the spot effect is at least comparable

to – if not much greater than – any true reflex Doppler signal from the companion. Similar

spot-induced RV amplitudes have already been observed for other T-Tauri stars in the optical

(e.g., Mahmud et al. 2011; Huerta et al. 2008; Prato et al. 2008). Since the star appears to be

co-rotating with the planet orbit, the period of such a spot-dominated RV signal would match the

orbital period, but the phase would be arbitrary, depending on the longitudinal spot placement.

Alternatively, spot and companion RV signals may both be significant and the two effects may

compete (see Section 3.3).

Regardless of the astrophysical cause behind the measured RV signal, we can place an upper

limit on the companion mass, and because of the good orbital phase coverage we can be reasonably

assured that possible aliasing effects are not a concern. We assume that any companion-induced

Doppler motion must be smaller than the amplitude of the measured variations, that Mp ≪ M∗,

and that M∗ = 0.39±0.10M⊙ (Briceno et al. 2005), and hence estimate Mp sin iorb ≤ 4.8±1.2MJup

(see, e.g., formula 3, Gaudi & Winn 2007). Taking our derived value of iorb we can directly estimate

Mp ≤ 5.5 ± 1.4MJup, comfortably within the planetary mass regime. The semi-amplitude (half

peak-to-peak) of the combined measured RV variations is ≈ 2.4 km s−1. This is separately confirmed

by the two individual datasets, which both have good phase coverage and show similar amplitudes

independent of the RV offset (as can be seen in Figure 9). By comparison, since Doppler-induced

RV semi-amplitude scales directly with Mp, a 25MJup object on the planet/brown-dwarf boundary

suggested by Schneider et al. (2011) would induce an 11 km s−1 signal; an object on the deuterium-

burning limit (≈ 13MJup) would induce a 5.7 km s−1 signal. It is unlikely, therefore, that the object

is of more than planetary mass, and still less in the stellar mass range.8 Since the spot distribution

is likely to change with time, further observations to look for changes in the phase of the RV signal

with respect to the transit ephemeris may provide more insight into the exact nature of the RV

signal. RV observations in the infra red are also likely to suffer less from spot-induced noise, and

so could also provide valuable further information.

We were unable to find any abnormalities in the observations regarding the outlying RV data

8We do note, however, that there is a possibility that if the reflex Doppler signature and the spot signature

are of comparable amplitudes, matching periods, and opposed in phase, they may destructively interfere, giving an

artificially low RV amplitude.

– 24 –

point, and, thus, have no reason to reject it as a bad measurement. However, falling at an orbital

phase φ = −0.082, it lies within the transit window (c.f. Figure 8), and its anomalous velocity

could reasonably be explained as arising from the Rossiter-McLaughlin (RM) effect (Rossiter 1924;

McLaughlin 1924). The RM effect would appear regardless of whether the main RV signal is caused

by spot modulation or stellar reflex motion, since it is a result of an asymmetric distortion of the

stellar absorption lines as an obscuration transits the unresolved stellar disk, blocking off regions

with successively different rotational red-shifts. Indeed, given the rapid stellar rotation, the RM

effect should be expected to appear during the transit window. From equation 6 of Gaudi & Winn

(2007) we can estimate the expected maximum amplitude of the effect, KR, given vr sin i∗, Rp, and

R∗. We find KR ≈ 3 km s−1, in good agreement with the offset of the outlier. The RM maximum

also lies typically around ∼ 1/5–1/3 of the way between transit-ingress and transit-center for gas-

giant planets, which is coincidentally where the outlier point is located. The sign of the offset of the

outlier data point, falling prior to the transit center time, is in agreement with a prograde orbit, as

we would expect if the star is in synchronous (or quasi-synchronous) rotation. It should be noted

that the RV data point at φ = +0.039 (HJD 2455671.755651) also lies within the transit window

but, in contrast, does not appear to be an outlier. Lying closer to the transit center time, one

would expect the RV offset to be smaller here; however, the exact form of the RM effect depends

on the precise orbital geometry, and may be partially masked by the choice of offset between the

two data sets. Further complication may also arise from the companion transiting photospheric

surface features. Dedicated RV measurements during transit could provide valuable confirmation of

the RM effect hinted at by the data, and help independently confirm the validity of the transiting

planet candidate.

3.2.2. Stellar Rotation

The stellar rotation velocity, vr, measured from the spectroscopy provides an independent

consistency check on the stellar rotation period. In order to estimate the projected rotation velocity,

vr sin i∗, we degrade the Kurucz synthetic model spectrum (see Section 2.4) to the resolution and

sampling of the HRS, and rotationally broaden it using a non-linear limb-darkening model (Claret

2000; Gray 1992) over the range of vr sin i∗ = 10 – 150 km s−1. We then Doppler shift the broadened

models, line by line, by cross-correlating against lines in our target spectrum. Subtracting a Doppler

shifted model from the observed spectrum gives an RMS residual for that rotational velocity.

Minimizing this residual gives the optimal rotational velocity measured for an individual line. We

applied this analysis to 12 lines that were sufficiently deep to be visible above the noise for our full

range of vr sin i∗ models. Weak lines could not be used because at large vr sin i∗ the line profiles

became buried in the noise. The weighted average of vr sin i∗ measured for these lines, across all

four HRS observations, was 80.6 ± 8.1 km s−1.

The assumed rotational period of the star and the radius measured from photometric transit

fitting (neglecting oblateness effects) independently imply an expected value for the unprojected

– 25 –

rotational velocity of vr = 2πR∗/P∗ ≈ 120 ± 11 km s−1. Taken together, the measured vr sin i∗ and

the photometrically derived vr give an estimated inclination of the stellar rotation axis, i∗ = 42±7◦.

Comparing this with the measured orbital inclination, iorb = 61.8 ± 3.7◦, we see that there is weak

evidence for a possible misalignment between the orbital and stellar rotation axes (consistent with

the possibility of detecting changes in the orbital inclination on relatively short timescales, as

mentioned in Section 3.1.2).

3.3. Implications

In Figure 10 we show the radius and mass of the companion in relation to the currently known

transiting exoplanets listed in the NASA Exoplanet Archive,9 where we have indicated the mass

of the candidate as an upper limit only. It clearly lies at the upper end of the gas-giant radius

distribution, although not unprecedentedly so. A significantly inflated atmosphere is to be expected

owing both to the system’s very young age, and to stellar irradiation at the companion’s exception-

ally small orbital radius. Planetary atmosphere models are currently not well constrained in this

regime due to the lack of known exoplanets that are both young and close-in. Marley et al. (2007)

and Fortney et al. (2007) caution that evolutionary models should be treated with care at ages

up to 10 Myr or more, being highly dependent on initial conditions and the formation mechanism

assumed. Indeed, Spiegel & Burrows (2012) suggest that comparison of atmospheric models with

observations of exoplanets at such young ages may be a good way of distinguishing between differ-

ent formation scenarios. Both Spiegel & Burrows (2012) and Marley et al. (2007) predict radii of

around 1.6RJup for a 5MJup planet at 3 Myr, for non-irradiated ‘hot-start’ (gravitational collapse)

model atmospheres; our error bars place our measured radius (1.9 ± 0.2R⊙) just above this value.

Our radius lies substantially further above the post-formation cold-start (core collapse) isochrones,

though comparison with the latter models is confused by uncertainty in the formation timescale,

which is likely comparable to the age of the star. Accounting for stellar irradiation could increase

the theoretical radius further. Increases of ∼ 10% or more are typical at ∼ 0.02 AU from a so-

lar like star (Baraffe et al. 2010; Chabrier et al. 2004), where the incident flux is similar to that

on our planet candidate (which is closer in, but orbiting a lower-mass star). The companion’s

proximity to the Roche limit may also weaken the gravitational binding of the object, further in-

creasing its expected radius. Various other explanations are also proposed to explain the excess

“radius anomaly” that is increasingly found in other gas-giant exoplanets (e.g., Baraffe et al. 2010;

Chabrier et al. 2011). Our estimated radius is therefore not unreasonable, but due to both the

model and observation uncertainties, a robust comparison with theory is difficult.

The apparent companion to PTFO 8-8695 also orbits close enough to its parent star that the

consequent small size of its Roche lobe may be relevant. Below a certain mass threshold, it may

not have sufficient self-gravity to hold itself together, and thus may begin to lose mass. Following

9http://exoplanetarchive.ipac.caltech.edu

– 26 –

Fig. 10.— Radius vs. mass for the known transiting planets (Section 3.3; data taken from the

NASA Exoplanet Archive). PTFO 8-8695b is marked as an upper mass limit, highlighted by the

large square. Iso-density contours are marked at 0.5, 1, 1.33 (Jupiter density), 5, and 10 g cm−3.

– 27 –

Faber et al. (2005) and Ford & Rasio (2006), the Roche radius, RRoche, according to Paczynski

(1971), is given by

RRoche = 0.462(Mp/M∗)1/3a ≤ 0.462(Mp,max/M∗)1/3a ≈ 1.92 ± 0.16, (1)

where Mp,max is our previously estimated maximum mass for the companion. Comparing with the

measured companion radius gives Rp/RRoche & 0.994 ± 0.094. Framing the argument another way,

we can rewrite the Roche formula to estimate the Roche limiting orbital radius, RRoche, in terms

of the measured companion radius to find a/aRoche . 1.008 ± 0.095 at the most – consistent with

being at or within the Roche limit, within the errors. The planet may be sufficiently inflated that

it fills its Roche lobe, and consequently may have lost mass in the past, or be in the process of

losing mass (thus maintaining itself at the Roche limit).

4. Conclusions

We have detected transits from a candidate young planet orbiting a previously-identified co-

rotating or near-co-rotating ≈ 2.7 Myr-old M3 weak-line T-Tauri star. Although we cannot com-

pletely rule out a false positive due to source blends, we are able to rule out confusion at the level

of ∆H ≈ 4.3 mag beyond a separation of 0.25′′, and argue qualitatively that a false detection due

to a blended eclipsing binary is unlikely. The companion is in an exceptionally rapid 0.448413 d

orbit, placing it among the shortest of the currently known exoplanet periods (cf. Demory et al.

2011; Charpinet et al. 2011; Muirhead et al. 2012; Rappaport et al. 2012). With an orbital radius

only around twice the stellar radius, it appears to be at or within its Roche limiting orbit, with

a/aRoche . 1.01±0.10, raising the possibility of past or ongoing evaporation and mass loss. Perhaps

the companion has been migrating, and losing any mass beyond its Roche lobe as it does so; or

perhaps it is continually being inflated to fill its Roche lobe, with any material which overflows

being stripped away.

Although the transit photometry and the RV data both phase-fold on the same periods, there

is an apparent offset in phase between them. The most likely explanation is that the RV signal is

shifted or dominated by the effect of star spots; we therefore suggest an upper limit on the (incli-

nation independent) companion mass of ≈ 5MJup based on the amplitude of the RV modulation.

If it can be assumed that the object has had time to reach a stable (or quasi-stable) state – i.e.,

that mass loss rates are not too rapid, and that there have been no recent dramatic changes in the

orbital geometry – then Roche limit considerations would imply a lower limit of a similar order,

since anything much less would be unable to gravitationally bind the material within the measured

companion radius.

The data are complex enough that we cannot yet be certain of a planet detection. In favor

of the planetary interpretation, however, we note that: (1) the photometric transit shape appears

to be flat bottomed with sharp ingress and egress slopes, and is difficult to explain with a star-

spot model alone; (2) the transits appear highly consistent and periodic over a period of & 1 yr,

– 28 –

which would be unusual for spots; (3) the RV signal places an upper mass limit well within the

planetary regime; (4) a false positive due to a faint blended eclipsing binary is argued against by

two observations which are suggestive of a transiting object associated with the primary T-Tauri

star observed: the stellar rotation rate appears very close to the transit period, suggesting co-

rotation; and the photometric variation from transit to transit is consistent with the notion of a

planet transiting a heavily spotted T-Tauri star.

Further spectroscopic and photometric observations are needed to provide a clearer picture.

Since the star is such a fast rotator, the expected RM effect (hinted at in the data) may provide a

valuable opportunity for confirmation of the candidate. Most RV noise sources will be constant in

the timescale of one transit, so further RV observations during the transit window could potentially

provide full sampling of the effect. This would help confirm the planetary interpretation of the data,

and could provide further useful information on the system geometry. Revisiting the target to make

further measurements of the RV phase with respect to the transit ephemeris would help confirm the

spot-interference interpretation: if spots are significant, the phase offset is likely to change with time

as the spots change. Infra-red RV observations, which are less sensitive to spot-induced noise, would

also provide valuable further information, perhaps allowing for an unambiguous determination of

the companion mass.

Given the young age of the system, the planet candidate is likely hot, and so a secondary

transit may also be detectable with more precise photometry, allowing constraints to be placed on

the companion temperature. Finally, simultaneous multi-band photometry and spectroscopy (par-

ticularly in the infra-red, where spot-effects should be lessened) could further help disentangle the

signatures of star spots and companion. This would also help to establish the cause of the apparent

overall change in transit shape between the two years’ observing runs, which could result either

from changes in the spot distribution or possibly, given the exceptionally short orbital timescale,

changes in the orbital geometry.

If our interpretation of the data is correct, the putative planet’s youth and its unique proximity

to its host star will make it a valuable object for helping inform our understanding of exoplanet

formation. Its young age could have important implications for the mechanism by which it formed:

conventional core-accretion formation models (Pollack et al. 1996) occur on timescales of ∼ 1–

10 Myr, comparable with or longer than the age of the system; the much more rapid gravitational-

instability mechanism (Boss 1997, 2000) occurs on timescales orders of magnitudes shorter, and

may thus perhaps be favored (see Baraffe et al. 2010, for a brief overview and comparison of the two

mechanisms). The companion’s inflated atmosphere appears indicative of the “hot start” models of

Marley et al. (2007) and Spiegel & Burrows (2012), which are associated with gravitational insta-

bility. At the same time, we cannot rule out that formation may yet be incomplete given the young

age of the system. Neither can we rule out that the companion may be on the verge of evaporation

and may not survive the lifetime of the star. Indeed, very few large planets are known to orbit

small stars, and this object may well be transient. Much ambiguity remains at these young ages,

and more observation and analysis of the system is needed. If the planetary nature of the proposed

– 29 –

companion is confirmed, the system could provide a wealth of new information for constraining

dynamical and atmospheric evolution models for exoplanets. We therefore propose PTFO 8-8695

as an object worthy of further careful study by the community.

The authors acknowledge Tim Morton and Dimitri Veras for their helpful input. S.B.C. wishes

to acknowledge generous support from Gary and Cynthia Bengier, the Richard and Rhoda Gold-

man Fund, National Aeronautics and Space Administration (NASA)/Swift grant NNX10AI21G,

NASA/Fermi grant NNX1OA057G, and National Science Foundation (NSF) grant AST–0908886.

S.M. acknowledges support from NSF awards AST-1006676, AST-1126413, PSARC, and the NASA

Astrobiology Institute.

Observations obtained with the Samuel Oschin Telescope at the Palomar Observatory as part

of the Palomar Transient Factory project, a scientific collaboration between the California Institute

of Technology, Columbia University, Las Cumbres Observatory, the Lawrence Berkeley National

Laboratory, the National Energy Research Scientific Computing Center, the University of Oxford,

and the Weizmann Institute of Science.

The Hobby-Eberly Telescope (HET) is a joint project of the University of Texas at Austin,

the Pennsylvania State University, Stanford University, Ludwig-Maximilians-Universitat Munchen,

and Georg-August-Universitat Gottingen. The HET is named in honor of its principal benefactors,

William P. Hobby and Robert E. Eberly.

Some of the data presented herein were obtained at the W.M. Keck Observatory, which is

operated as a scientific partnership among the California Institute of Technology, the University of

California and the National Aeronautics and Space Administration. The Observatory was made

possible by the generous financial support of the W.M. Keck Foundation. The authors wish to

recognize and acknowledge the very significant cultural role and reverence that the summit of

Mauna Kea has always had within the indigenous Hawaiian community. We are most fortunate to

have the opportunity to conduct observations from this mountain.

The Byrne Observatory at Sedgwick (BOS) is operated by the Las Cumbres Observatory Global

Telescope Network and is located at the Sedgwick Reserve, a part of the University of California

Natural Reserve System.

This research has made use of the NASA Exoplanet Archive, which is operated by the California

Institute of Technology, under contract with the National Aeronautics and Space Administration

under the Exoplanet Exploration Program.

This publication makes use of data products from the Two Micron All Sky Survey, which

is a joint project of the University of Massachusetts and the Infrared Processing and Analysis

Center/California Institute of Technology, funded by the National Aeronautics and Space Admin-

istration and the National Science Foundation.

This research has made use of the VizieR catalogue access tool, CDS, Strasbourg, France; and

– 30 –

of the software package Uncertainties: a Python package for calculations with uncertainties, Eric

O. Lebigot.10

This work was partially supported by funding from the Center for Exoplanets and Habitable

Worlds. The Center for Exoplanets and Habitable Worlds is supported by the Pennsylvania State

University, the Eberly College of Science, and the Pennsylvania Space Grant Consortium.

Support for this work was provided by an award issued by JPL/Caltech.

Facilities: PO:1.2m (PTF), HET (HRS), Keck:I (HIRES), Keck:II (NIRC2), LCOGT (BOS),

FTN, FTS, Mayall, Hale

REFERENCES

Agueros, M. A., Covey, K. R., Lemonias, J. J., Law, N. M., Kraus, A., Batalha, N., Bloom, J. S.,

Cenko, S. B., Kasliwal, M. M., Kulkarni, S. R., Nugent, P. E., Ofek, E. O., Poznanski, D.,

& Quimby, R. M. 2011, ApJ, 740, 110

Aigrain, S., Hodgkin, S., Irwin, J., Hebb, L., Irwin, M., Favata, F., Moraux, E., & Pont, F. 2007,

MNRAS, 375, 29

Armitage, P. J. 2009, in American Institute of Physics Conference Series, Vol. 1192, American

Institute of Physics Conference Series, ed. F. Roig, D. Lopes, R. de La Reza, & V. Ortega,

3–42

Baraffe, I., Chabrier, G., Allard, F., & Hauschildt, P. H. 1998, A&A, 337, 403

Baraffe, I., Chabrier, G., & Barman, T. 2010, Reports on Progress in Physics, 73, 016901

Bender, C. F., Mahadevan, S., Deshpande, R., Wright, J. T., Roy, A., Terrien, R. C., Sigurdsson,

S., Ramsey, L. W., Schneider, D. P., & Fleming, S. W. 2012, ApJ, 751, L31

Boss, A. P. 1997, Science, 276, 1836

—. 2000, ApJ, 536, L101

Briceno, C., Calvet, N., Hernandez, J., Vivas, A. K., Hartmann, L., Downes, J. J., & Berlind, P.

2005, AJ, 129, 907

Briceno, C., Hartmann, L., Hernandez, J., Calvet, N., Vivas, A. K., Furesz, G., & Szentgyorgyi, A.

2007, ApJ, 661, 1119

Carter, J. A. & Winn, J. N. 2009, ApJ, 704, 51

10http://packages.python.org/uncertainties

– 31 –

Chabrier, G., Barman, T., Baraffe, I., Allard, F., & Hauschildt, P. H. 2004, ApJ, 603, L53

Chabrier, G., Leconte, J., & Baraffe, I. 2011, in IAU Symposium, Vol. 276, IAU Symposium, ed.

A. Sozzetti, M. G. Lattanzi, & A. P. Boss, 171–180

Charpinet, S., Fontaine, G., Brassard, P., Green, E. M., van Grootel, V., Randall, S. K., Silvotti,

R., Baran, A. S., Østensen, R. H., Kawaler, S. D., & Telting, J. H. 2011, Nature, 480, 496

Claret, A. 2000, A&A, 363, 1081

Crockett, C. J., Mahmud, N. I., Prato, L., Johns-Krull, C. M., Jaffe, D. T., & Beichman, C. A.

2011, ApJ, 735, 78

Dawson, R. I. & Fabrycky, D. C. 2010, ApJ, 722, 937

Demory, B.-O., Gillon, M., Deming, D., Valencia, D., Seager, S., Benneke, B., Lovis, C., Cubillos,

P., Harrington, J., Stevenson, K. B., Mayor, M., Pepe, F., Queloz, D., Segransan, D., &

Udry, S. 2011, A&A, 533, A114

Esposito, M., Guenther, E., Hatzes, A. P., & Hartmann, M. 2006, in Tenth Anniversary of 51 Peg-b:

Status of and prospects for hot Jupiter studies, ed. L. Arnold, F. Bouchy, & C. Moutou,

127–134

Faber, J. A., Rasio, F. A., & Willems, B. 2005, Icarus, 175, 248

Ford, E. B. & Rasio, F. A. 2006, ApJ, 638, L45

Fortney, J. J., Marley, M. S., & Barnes, J. W. 2007, ApJ, 659, 1661

Fortney, J. J. & Nettelmann, N. 2010, Space Sci. Rev., 152, 423

Gaudi, B. S. & Winn, J. N. 2007, ApJ, 655, 550

Gazak, J. Z., Johnson, J. A., Tonry, J., Dragomir, D., Eastman, J., Mann, A. W., & Agol, E. 2012,

Advances in Astronomy, 2012, 697967

Gray, D. F. 1992, The Observation and Analysis of Stellar Photospheres, 2nd edn. (Cambridge;

New York: Cambridge University Press)

Heacox, W. D. 1987, Journal of the Optical Society of America A, 4, 488

Hernandez, J., Calvet, N., Briceno, C., Hartmann, L., Vivas, A. K., Muzerolle, J., Downes, J.,

Allen, L., & Gutermuth, R. 2007, ApJ, 671, 1784

Hillenbrand, L. A. 2008, Physica Scripta Volume T, 130, 014024

Howard, A. W., Johnson, J. A., Marcy, G. W., Fischer, D. A., Wright, J. T., Henry, G. W.,

Giguere, M. J., Isaacson, H., Valenti, J. A., Anderson, J., & Piskunov, N. E. 2009, ApJ,

696, 75

– 32 –

Huerta, M., Johns-Krull, C. M., Prato, L., Hartigan, P., & Jaffe, D. T. 2008, ApJ, 678, 472

Kraus, A. L. & Ireland, M. J. 2012, ApJ, 745, 5

Kurucz, R. L. 2005, Memorie della Societa Astronomica Italiana Supplementi, 8, 14

Law, N. M., Dekany, R. G., Rahmer, G., Hale, D., Smith, R., Quimby, R., Ofek, E. O., Kasliwal,

M., Zolkower, J., Velur, V., Henning, J., Bui, K., McKenna, D., Nugent, P., Jacobsen, J.,

Walters, R., Bloom, J., Surace, J., Grillmair, C., Laher, R., Mattingly, S., & Kulkarni,

S. 2010, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series,

Vol. 7735, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series,

77353M

Law, N. M., Kulkarni, S. R., Dekany, R. G., Ofek, E. O., Quimby, R. M., Nugent, P. E., Surace,

J., Grillmair, C. C., Bloom, J. S., Kasliwal, M. M., Bildsten, L., Brown, T., Cenko, S. B.,

Ciardi, D., Croner, E., Djorgovski, S. G., van Eyken, J., Filippenko, A. V., Fox, D. B., Gal-

Yam, A., Hale, D., Hamam, N., Helou, G., Henning, J., Howell, D. A., Jacobsen, J., Laher,

R., Mattingly, S., McKenna, D., Pickles, A., Poznanski, D., Rahmer, G., Rau, A., Rosing,

W., Shara, M., Smith, R., Starr, D., Sullivan, M., Velur, V., Walters, R., & Zolkower, J.

2009, PASP, 121, 1395

Levitan, D., Fulton, B. J., Groot, P. J., Kulkarni, S. R., Ofek, E. O., Prince, T. A., Shporer, A.,

Bloom, J. S., Cenko, S. B., Kasliwal, M. M., Law, N. M., Nugent, P. E., Poznanski, D.,

Quimby, R. M., Horesh, A., Sesar, B., & Sternberg, A. 2011, ApJ, 739, 68

Lomb, N. R. 1976, Ap&SS, 39, 447

Mahmud, N. I., Crockett, C. J., Johns-Krull, C. M., Prato, L., Hartigan, P. M., Jaffe, D. T., &

Beichman, C. A. 2011, ApJ, 736, 123

Mandel, K. & Agol, E. 2002, ApJ, 580, L171

Marley, M. S., Fortney, J. J., Hubickyj, O., Bodenheimer, P., & Lissauer, J. J. 2007, ApJ, 655, 541

Marsh, K. A., Kirkpatrick, J. D., & Plavchan, P. 2010, ApJ, 709, L158

McLaughlin, D. B. 1924, ApJ, 60, 22

Miller, A. A., Irwin, J., Aigrain, S., Hodgkin, S., & Hebb, L. 2008, MNRAS, 387, 349

Miralda-Escude, J. 2002, ApJ, 564, 1019

Muirhead, P. S., Johnson, J. A., Apps, K., Carter, J. A., Morton, T. D., Fabrycky, D. C., Pineda,

J. S., Bottom, M., Rojas-Ayala, B., Schlawin, E., Hamren, K., Covey, K. R., Crepp, J. R.,

Stassun, K. G., Pepper, J., Hebb, L., Kirby, E. N., Howard, A. W., Isaacson, H. T., Marcy,

G. W., Levitan, D., Diaz-Santos, T., Armus, L., & Lloyd, J. P. 2012, ApJ, 747, 144

– 33 –

Neuhauser, R., Errmann, R., Berndt, A., Maciejewski, G., Takahashi, H., Chen, W. P., Dimitrov,

D. P., Pribulla, T., Nikogossian, E. H., Jensen, E. L. N., Marschall, L., Wu, Z.-Y., Kellerer,

A., Walter, F. M., Briceno, C., Chini, R., Fernandez, M., Raetz, S., Torres, G., Latham,

D. W., Quinn, S. N., Niedzielski, A., Bukowiecki, L., Nowak, G., Tomov, T., Tachihara, K.,

Hu, S. C.-L., Hung, L. W., Kjurkchieva, D. P., Radeva, V. S., Mihov, B. M., Slavcheva-

Mihova, L., Bozhinova, I. N., Budaj, J., Vanko, M., Kundra, E., Hambalek, L., Krushevska,

V., Movsessian, T., Harutyunyan, H., Downes, J. J., Hernandez, J., Hoffmeister, V. H.,

Cohen, D. H., Abel, I., Ahmad, R., Chapman, S., Eckert, S., Goodman, J., Guerard, A.,

Kim, H. M., Koontharana, A., Sokol, J., Trinh, J., Wang, Y., Zhou, X., Redmer, R., Kramm,

U., Nettelmann, N., Mugrauer, M., Schmidt, J., Moualla, M., Ginski, C., Marka, C., Adam,

C., Seeliger, M., Baar, S., Roell, T., Schmidt, T. O. B., Trepl, L., Eisenbeiß, T., Fiedler, S.,

Tetzlaff, N., Schmidt, E., Hohle, M. M., Kitze, M., Chakrova, N., Grafe, C., Schreyer, K.,

Hambaryan, V. V., Broeg, C. H., Koppenhoefer, J., & Pandey, A. K. 2011, Astronomische

Nachrichten, 332, 547

Nguyen, D. C., Brandeker, A., van Kerkwijk, M. H., & Jayawardhana, R. 2012, ApJ, 745, 119

Ofek, E. O., Laher, R., Law, N., Surace, J., Levitan, D., Sesar, B., Horesh, A., Poznanski, D., van

Eyken, J. C., Kulkarni, S. R., Nugent, P., Zolkower, J., Walters, R., Sullivan, M., Agueros,

M., Bildsten, L., Bloom, J., Cenko, S. B., Gal-Yam, A., Grillmair, C., Helou, G., Kasliwal,

M. M., & Quimby, R. 2012, PASP, 124, 62

Paczynski, B. 1971, ARA&A, 9, 183

Paulson, D. B. & Yelda, S. 2006, PASP, 118, 706

Plavchan, P., Jura, M., Kirkpatrick, J. D., Cutri, R. M., & Gallagher, S. C. 2008, ApJS, 175, 191

Pollack, J. B., Hubickyj, O., Bodenheimer, P., Lissauer, J. J., Podolak, M., & Greenzweig, Y. 1996,

Icarus, 124, 62

Prato, L., Huerta, M., Johns-Krull, C. M., Mahmud, N., Jaffe, D. T., & Hartigan, P. 2008, ApJ,

687, L103

Ramsey, L. W., Adams, M. T., Barnes, T. G., Booth, J. A., Cornell, M. E., Fowler, J. R., Gaffney,

N. I., Glaspey, J. W., Good, J. M., Hill, G. J., Kelton, P. W., Krabbendam, V. L., Long, L.,

MacQueen, P. J., Ray, F. B., Ricklefs, R. L., Sage, J., Sebring, T. A., Spiesman, W. J., &

Steiner, M. 1998, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference

Series, Vol. 3352, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference

Series, ed. L. M. Stepp, 34–42

Rappaport, S., Levine, A., Chiang, E., El Mellah, I., Jenkins, J., Kalomeni, B., Kite, E. S., Kotson,

M., Nelson, L., Rousseau-Nepton, L., & Tran, K. 2012, ApJ, 752, 1

– 34 –

Rau, A., Kulkarni, S. R., Law, N. M., Bloom, J. S., Ciardi, D., Djorgovski, G. S., Fox, D. B.,

Gal-Yam, A., Grillmair, C. C., Kasliwal, M. M., Nugent, P. E., Ofek, E. O., Quimby, R. M.,

Reach, W. T., Shara, M., Bildsten, L., Cenko, S. B., Drake, A. J., Filippenko, A. V., Helfand,

D. J., Helou, G., Howell, D. A., Poznanski, D., & Sullivan, M. 2009, PASP, 121, 1334

Rossiter, R. A. 1924, ApJ, 60, 15

Scargle, J. D. 1982, ApJ, 263, 835

Schneider, J., Dedieu, C., Le Sidaner, P., Savalle, R., & Zolotukhin, I. 2011, A&A, 532, A79

Setiawan, J., Henning, T., Launhardt, R., Muller, A., Weise, P., & Kurster, M. 2008, Nature, 451,

38

Setiawan, J., Weise, P., Henning, T., Launhardt, R., Muller, A., & Rodmann, J. 2007, ApJ, 660,

L145

Shetrone, M., Cornell, M. E., Fowler, J. R., Gaffney, N., Laws, B., Mader, J., Mason, C., Odewahn,

S., Roman, B., Rostopchin, S., Schneider, D. P., Umbarger, J., & Westfall, A. 2007, PASP,

119, 556

Siess, L., Dufour, E., & Forestini, M. 2000, A&A, 358, 593

Skrutskie, M. F., Cutri, R. M., Stiening, R., Weinberg, M. D., Schneider, S., Carpenter, J. M.,

Beichman, C., Capps, R., Chester, T., Elias, J., Huchra, J., Liebert, J., Lonsdale, C.,

Monet, D. G., Price, S., Seitzer, P., Jarrett, T., Kirkpatrick, J. D., Gizis, J. E., Howard,

E., Evans, T., Fowler, J., Fullmer, L., Hurt, R., Light, R., Kopan, E. L., Marsh, K. A.,

McCallon, H. L., Tam, R., Van Dyk, S., & Wheelock, S. 2006, AJ, 131, 1163

Spiegel, D. S. & Burrows, A. 2012, ApJ, 745, 174

Tull, R. G. 1998, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series,

Vol. 3355, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series,

ed. S. D’Odorico, 387–398

van Eyken, J. C., Ciardi, D. R., Rebull, L. M., Stauffer, J. R., Akeson, R. L., Beichman, C. A.,

Boden, A. F., von Braun, K., Gelino, D. M., Hoard, D. W., Howell, S. B., Kane, S. R.,

Plavchan, P., Ramırez, S. V., Bloom, J. S., Cenko, S. B., Kasliwal, M. M., Kulkarni, S. R.,

Law, N. M., Nugent, P. E., Ofek, E. O., Poznanski, D., Quimby, R. M., Grillmair, C. J.,

Laher, R., Levitan, D., Mattingly, S., & Surace, J. A. 2011, AJ, 142, 60

Vogt, S. S., Allen, S. L., Bigelow, B. C., Bresee, L., Brown, B., Cantrall, T., Conrad, A., Couture,

M., Delaney, C., Epps, H. W., Hilyard, D., Hilyard, D. F., Horn, E., Jern, N., Kanto, D.,

Keane, M. J., Kibrick, R. I., Lewis, J. W., Osborne, J., Pardeilhan, G. H., Pfister, T., Rick-

etts, T., Robinson, L. B., Stover, R. J., Tucker, D., Ward, J., & Wei, M. Z. 1994, in Society

of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 2198, Society of

– 35 –

Photo-Optical Instrumentation Engineers (SPIE) Conference Series, ed. D. L. Crawford &

E. R. Craine, 362

Wright, J. T. & Howard, A. W. 2009, ApJS, 182, 205

This preprint was prepared with the AAS LATEX macros v5.2.